At the start of their Fall 2017 Art of Writing seminar, “Communicating with Data,” Professor Deborah Nolan and then-graduate student Sara Stoudt showed their undergraduates a series of articles from the popular press.
The articles were written by scientists who had made a point of presenting the results of their data analysis to the general public in an accessible, approachable manner. Their prose communicated the specificities and subtleties of their findings without veering into technical jargon; each managed the delicate task of balancing readability and rigor.
The point, Nolan explained, was for her data science students to understand “the power of communicating beyond their circle, beyond people who know exactly what they know.” The articles underscored the critical importance of written communication to the practice of data science and provided students with models for the kinds of writing they would be learning to produce.
Nolan, now a professor emeritus in the Department of Statistics, had been teaching theoretical statistics at Berkeley for over two decades before she encountered the university’s Art of Writing program.
In her courses, she asked her students to practice writing up their findings for different audiences — to pretend they had to translate their data analysis into a memo to a boss, or a publicly-distributed pamphlet, or a press release. She emphasized the outsize role that writing played in the field, but had not yet figured out how to teach writing for statistics to her students.
So, when she saw that Art of Writing was soliciting proposals for its seminar development grant and that it was particularly seeking applications from faculty outside the traditional fields of writing, “I jumped on it,” she said. “I had been wanting to teach students this kind of writing for 25 years.”
Nolan asked the graduate students in her department if one of them would be interested in teaming up with her to write the proposal and teach the course, and Stoudt raised her hand. “She brought it to me and said, ‘Hey, want to do this class?’” Stoudt recalled. “As people working in quantitative spaces, we are not trained as writers and yet that’s what we spend so much of our time doing,” she explained.
Nolan’s invitation encouraged Stoudt to shift her thinking about her own professional identity: “Yes, I’m a statistician, but I’m also a writer,” she said.
The two spent the summer developing the course and planning activities for their students to take on. The main objective of the course was to have students produce a written report of a statistical project. Mini-exercises helped them along the way, designed to answer questions such as: How do you write captions for a figure? How do you edit someone else’s work? How do you come up with a title for your article? Students would also Tweet and blog their findings, and revise their prose multiple times.
“The act of communicating your findings is a way to really learn about statistics in a deeper, more nuanced sense,” Nolan said. It is impossible to be an effective statistician without knowing how to use words to convey analytical meaning. Yet before they created it, not one course was designed to teach students how to do exactly that.
After the course concluded, Stoudt and Nolan continued their collaboration. They began pushing for a communications course to be a permanent elective for the statistics major, and started expanding their course materials into a series of articles and a book.
“At the end of it, we said, ‘Let’s keep going.’ It was something that we both thought the data science, statistics, and the scientific community more broadly could really use,” Nolan said.
An Art of Writing Manuscript Seminar enabled them to convene a diverse array of faculty and data science experts to read their work-in-progress and provide critical feedback on how their course materials should be expanded into a textbook.
“Everybody read a different piece of the book; we spent half a day going through the content,” Nolan recalled. “That led us to do a major reorganization of the book, to broaden it to be for general science writing.”
Communicating with Data: The Art of Writing for Data Science was published by Oxford University Press in fall 2021. Professor Emeritus David Aldous gave it an “unsolicited enthusiastic recommendation” on his blog: “You might fear that a 300 page textbook on “technical writing” would be rather dry and sterile. But no! It offers a wide scope, from conceptual overviews of organization, to presentation of graphics and pseudo-code, to detailed analyses of individual sentences and words. A recurrent theme is that one should learn how to write by critically reading what others have written.”
York University, in Toronto, plans to use it as a textbook for a required communications course in its brand-new data science major. One of Nolan’s projects for the next year is to help Berkeley design a similar required communications class for its data science majors. In Spring 2018, Nolan and Stoudt co-taught a follow-up course, “Blogging for Data Science,” in which students repeatedly revised blog posts about their work and ultimately published them on the course website.
Data science is still a brand-new field, Nolan explained, and it is important for students to have a space to engage with critical questions about the impacts of their work, like, “What’s the human context around working with data? What are the ethics of working with data?”
The pandemic has further driven this point home: “Especially now, the broader public has to deal with numbers every day,” Stoudt said. “With the pandemic, it’s numbers all the time — that has been a sign that we cannot forget about it or neglect it.” Stoudt is now an assistant professor in the Mathematics Department at Bucknell University.
Both agreed that the art of writing for data science is not, at the end of the day, so different from the art of writing writ large.
“The notion that there is an argument to make, as opposed to a cut-and-dried conveyance of the statistical output–that’s the art,” Nolan said. “You’re still making an argument, and to make that argument, you need to take what you’ve done, put it in context, and show that you are aware of others’ work. You need to write it in a way that’s engaging, yet truthful and faithful to what the data shows.”